| { | |
| "best_metric": 3.17564058303833, | |
| "best_model_checkpoint": "./snap_diff_llama/diff_llama_410m_diff_attn/checkpoint-14000", | |
| "epoch": 0.024642857142857143, | |
| "eval_steps": 1000, | |
| "global_step": 14000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.1428571428571426e-07, | |
| "loss": 11.0205, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00021428571428571427, | |
| "loss": 6.3862, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "eval_loss": 4.933394432067871, | |
| "eval_ppl": 138.85003031304973, | |
| "eval_runtime": 32.6853, | |
| "eval_samples_per_second": 15.297, | |
| "eval_steps_per_second": 0.122, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0002983246239337692, | |
| "loss": 4.4203, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "eval_loss": 4.056069374084473, | |
| "eval_ppl": 57.74688300953943, | |
| "eval_runtime": 31.9488, | |
| "eval_samples_per_second": 15.65, | |
| "eval_steps_per_second": 0.125, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00028822143178056114, | |
| "loss": 3.8205, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_loss": 3.6541380882263184, | |
| "eval_ppl": 38.63420748222174, | |
| "eval_runtime": 31.7672, | |
| "eval_samples_per_second": 15.74, | |
| "eval_steps_per_second": 0.126, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0002695698760834384, | |
| "loss": 3.5676, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "eval_loss": 3.523311138153076, | |
| "eval_ppl": 33.896478778578945, | |
| "eval_runtime": 33.1855, | |
| "eval_samples_per_second": 15.067, | |
| "eval_steps_per_second": 0.121, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00024352347027881003, | |
| "loss": 3.4614, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "eval_loss": 3.4199113845825195, | |
| "eval_ppl": 30.566706219598544, | |
| "eval_runtime": 32.8838, | |
| "eval_samples_per_second": 15.205, | |
| "eval_steps_per_second": 0.122, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00021169306546959174, | |
| "loss": 3.3952, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 3.3484232425689697, | |
| "eval_ppl": 28.457827159310067, | |
| "eval_runtime": 32.4291, | |
| "eval_samples_per_second": 15.418, | |
| "eval_steps_per_second": 0.123, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00017604722665003956, | |
| "loss": 3.347, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "eval_loss": 3.274552822113037, | |
| "eval_ppl": 26.431403298374143, | |
| "eval_runtime": 32.6886, | |
| "eval_samples_per_second": 15.296, | |
| "eval_steps_per_second": 0.122, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00013879048596203636, | |
| "loss": 3.2857, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "eval_loss": 3.2431113719940186, | |
| "eval_ppl": 25.613290373707557, | |
| "eval_runtime": 32.5175, | |
| "eval_samples_per_second": 15.376, | |
| "eval_steps_per_second": 0.123, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00010222700246224735, | |
| "loss": 3.2699, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_loss": 3.204813003540039, | |
| "eval_ppl": 24.6508899562805, | |
| "eval_runtime": 32.063, | |
| "eval_samples_per_second": 15.594, | |
| "eval_steps_per_second": 0.125, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 6.86180604201361e-05, | |
| "loss": 3.2255, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "eval_loss": 3.1874945163726807, | |
| "eval_ppl": 24.227649356846815, | |
| "eval_runtime": 31.9699, | |
| "eval_samples_per_second": 15.64, | |
| "eval_steps_per_second": 0.125, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.004221922552608e-05, | |
| "loss": 3.2335, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "eval_loss": 3.1818697452545166, | |
| "eval_ppl": 24.091756914706085, | |
| "eval_runtime": 32.8207, | |
| "eval_samples_per_second": 15.234, | |
| "eval_steps_per_second": 0.122, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.82667639944657e-05, | |
| "loss": 3.2553, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "eval_loss": 3.178798198699951, | |
| "eval_ppl": 24.017871491109133, | |
| "eval_runtime": 32.5604, | |
| "eval_samples_per_second": 15.356, | |
| "eval_steps_per_second": 0.123, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1e-05, | |
| "loss": 3.237, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "eval_loss": 3.1790049076080322, | |
| "eval_ppl": 24.022836712259533, | |
| "eval_runtime": 32.1732, | |
| "eval_samples_per_second": 15.541, | |
| "eval_steps_per_second": 0.124, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1e-05, | |
| "loss": 3.227, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "eval_loss": 3.17564058303833, | |
| "eval_ppl": 23.94215189353734, | |
| "eval_runtime": 32.4234, | |
| "eval_samples_per_second": 15.421, | |
| "eval_steps_per_second": 0.123, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "step": 14000, | |
| "total_flos": 2.0002685775324905e+19, | |
| "train_loss": 0.07952269635881697, | |
| "train_runtime": 1357.4941, | |
| "train_samples_per_second": 2640.159, | |
| "train_steps_per_second": 10.313 | |
| } | |
| ], | |
| "logging_steps": 1000, | |
| "max_steps": 14000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 1000, | |
| "total_flos": 2.0002685775324905e+19, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |